萬盛學電腦網

您现在的位置: 萬盛學電腦網 >> 程式設計 >  >> 網路程式設計 >> PHP 程式設計

PHP 並行 HTTP 請求:cURL Multi 介面封裝類 (HttpMultiRequest)

動機

須同時抓取多個 API/網頁,串行請求總耗時 = Σ(單次耗時);並行請求總耗時 ≈ max(單次耗時),大幅提升效能。

封裝類別

<?php
class HttpMultiRequest {
    private $urls = [];
    private $options = [];

    public function __construct(array $options = []) {
        $this->setOptions($options);
    }

    public function setUrls(array $urls): self {
        $this->urls = $urls; return $this;
    }

    public function setOptions(array $opts): self {
        $defaults = [
            CURLOPT_RETURNTRANSFER => true,
            CURLOPT_FOLLOWLOCATION => true,
            CURLOPT_MAXREDIRS      => 5,
            CURLOPT_CONNECTTIMEOUT => 10,
            CURLOPT_TIMEOUT        => 30,
            CURLOPT_USERAGENT      => 'Mozilla/5.0 (compatible; PHP HttpMultiRequest)',
            CURLOPT_SSL_VERIFYPEER => true,
            CURLOPT_SSL_VERIFYHOST => 2,
        ];
        $this->options = $defaults + $opts;
        return $this;
    }

    /** 並行執行所有 URL,回傳 [url => body] */
    public function execute(): array {
        if (empty($this->urls)) return [];

        $mh = curl_multi_init();
        $handles = [];
        foreach ($this->urls as $i => $url) {
            $ch = curl_init($url);
            curl_setopt_array($ch, $this->options);
            curl_multi_add_handle($mh, $ch);
            $handles[$i] = $ch;
        }

        // 執行多工迴圈 (正確寫法:使用 curl_multi_select 避免 CPU 100%)
        $running = null;
        do {
            curl_multi_exec($mh, $running);
            if ($running) {
                // 等待活動,最長 1 秒
                curl_multi_select($mh, 1.0);
            }
        } while ($running > 0);

        // 讀取結果
        $results = [];
        foreach ($handles as $i => $ch) {
            $results[$this->urls[$i]] = curl_multi_getcontent($ch);
            curl_multi_remove_handle($mh, $ch);
        }
        curl_multi_close($mh);
        return $results;
    }

    /** 單一請求 (串行) */
    public function executeOne(string $url): string|false {
        $ch = curl_init($url);
        curl_setopt_array($ch, $this->options);
        $content = curl_exec($ch);
        curl_close($ch);
        return $content;
    }
}

// 測試對比
$urls = array_fill(0, 8, 'https://httpbin.org/delay/1'); // 每請求延遲 1 秒

$multi = new HttpMultiRequest();
$start = microtime(true);
$resM = $multi->setUrls($urls)->execute();
echo "Parallel: " . (microtime(true) - $start) . "s\n"; // ~1.2s

$start = microtime(true);
$resS = [];
foreach ($urls as $u) $resS[] = $multi->executeOne($u);
echo "Serial: " . (microtime(true) - $start) . "s\n";   // ~8s

關鍵修正:避免 CPU 100% 的正確多工迴圈

原文採用 do { curl_multi_exec($mh,$running); } while($running>0); **會導致 CPU 100% 空轉**。正確做法需呼叫 curl_multi_select($mh, $timeout) 讓進程睡眠等待 socket 事件。

進階功能建議

  • 錯誤處理:檢查 curl_errno($ch)、curl_getinfo($ch, CURLINFO_HTTP_CODE),失敗可重試。
  • 並發限制:大量 URL 時分批 (如每批 50 個),避免耗盡檔案描述符或觸發對方限流。
  • 超時控制:整體逾時、單次連線/傳輸逾時分開設定。
  • 現代替代:GuzzleHttp\Pool、amphp/http-client (非阻塞)、react/http-client。

適用場景

  1. 聚合多個微服務 API 回應
  2. 爬蟲並行抓取列表頁
  3. 健康檢查多個端點
  4. 批次推送 Webhook
copyright © 萬盛學電腦網 all rights reserved